Nonparametric comparisons with \(k_2>2\) controls using normal scores and Savage statistics (Q2463657): Difference between revisions

The subject of this paper is to determine which treatments are better than existing controls. The author develops distribution-free procedures for simultaneously comparing \(k_1\) treatments to \(k_2\) controls using normal scores and \textit{I.R. Savage} statistics [see Ann. Math. Stat. 27, 590--615 (1956; Zbl 0073.13904)]. A method for computation of the upper bounds for the type I error probabilities is shown. The author also presents an example using the breaking strength of fibers.